Field value pairs (File architecture)
Overview
The basic unit of storage inside a Model 204 record is the field value pair (or, more literally, "field = value pair"). A record or a fieldgroup contains a series of fieldgroups and field value pairs.
A record can contain field value pairs for fields with the OCCURS attribute, which are described ??.
A record or a fieldgroup can contain non-OCCURS field value pairs, whose physical structure is described below, for each storage data type.
String Fields
The field value pair for string fields contains three parts:
- the field code from Table A. This is 3 bytes for Enhanced Data Handling Files (FILEORG x'100' files) and 2 bytes for all others.
- 1 byte containing the length of the value for this pair (0 to 255).
- the value. Held in the number of bytes above.
So, if you have a NAME field containing 'SMITH' it will be held in either 8 or 9 bytes (depending on the FILEORG), while a NAME of 'FRANKENSTEIN' will be held as either 15 or 16.
Binary Fields
Binary fields are stored (and work) differently depending on whether the file has the Optimized Field Extraction (FILEORG x'80') bit set.
For non Optimized Field Extraction Files the field value pair contains two parts:
- the field code from Table A. This is 2 bytes. (Enhanced Data Handling Files (FILEORG x'100' files automatically also turn on x'80'.)
- the value which is always held as 4 bytes.
For Optimized Field Extraction Files the field value pair contains three parts:
- the field code from Table A. This is 3 bytes for Optimized Field Extraction Files (FILEORG x'100' files) and 2 bytes for all others.
- one byte containing the length of the value (1 to 4 bytes).
- the value. Its length is as above.
This effectively makes the BINARY field storage similar to string fields, and improves the speed with which the data in the record can be read. The possible tradeoff on performance versus space is discussed in File Design.
Float Fields
Float fields consist of two parts: